Mechanism of H₂S-mediated protection against oxidative stress in Escherichia coli

Significance

Hydrogen sulfide (H₂S) is a highly toxic gas that interferes with cellular respiration; however, at low physiological amounts, it plays an important role in cell signaling. Remarkably, in bacteria, endogenously produced H₂S has been recently recognized as a general protective molecule, which renders multiple bacterial species highly resistant to oxidative stress and various classes of antibiotics. The mechanism of this phenomenon remains poorly understood. In this paper, we use Escherichia coli as a model system to elucidate its major enzymatic source of H₂S and establish the principle biochemical pathways that account for H₂S-mediated protection against reactive oxygen species. Understanding those mechanisms has far-reaching implications in preventing bacterial resistance and designing effective antimicrobial therapies.

Abstract

Endogenous hydrogen sulfide (H₂S) renders bacteria highly resistant to oxidative stress, but its mechanism remains poorly understood. Here, we report that 3-mercaptopyruvate sulfurtransferase (3MST) is the major source of endogenous H₂S in Escherichia coli. Cellular resistance to H₂O₂ strongly depends on the activity of mstA, a gene that encodes 3MST. Deletion of the ferric uptake regulator (Fur) renders ∆mstA cells hypersensitive to H₂O₂. Conversely, induction of chromosomal mstA from a strong pLtetO-1 promoter (P_tet-mstA) renders ∆fur cells fully resistant to H₂O₂. Furthermore, the endogenous level of H₂S is reduced in ∆fur or ∆sodA ∆sodB cells but restored after the addition of an iron chelator dipyridyl. Using a highly sensitive reporter of the global response to DNA damage (SOS) and the TUNEL assay, we show that 3MST-derived H₂S protects chromosomal DNA from oxidative damage. We also show that the induction of the CysB regulon in response to oxidative stress depends on 3MST, whereas the CysB-regulated l-cystine transporter, TcyP, plays the principle role in the 3MST-mediated generation of H₂S. These findings led us to propose a model to explain the interplay between l-cysteine metabolism, H₂S production, and oxidative stress, in which 3MST protects E. coli against oxidative stress via l-cysteine utilization and H₂S-mediated sequestration of free iron necessary for the genotoxic Fenton reaction.

Hydrogen sulfide (H₂S) is well-recognized as a second messenger implicated in many physiological processes in mammals, including synaptic transmission, vascular tone, inflammation, angiogenesis, and protection from oxidative stress (1). The latter function of H₂S seems to be universal, because it has been implicated in bacterial defense against reactive oxygen species (ROS) and antibiotics-induced oxidative damage (2). H₂S can also kill microorganisms by inhibiting antioxidant enzymes during induced oxidative stress (3, 4). These seemingly contradictory attributes of H₂S highlight its concentration-dependent dual nature: at high concentration, it is a toxic gas, and at lower physiological concentrations, it is a signaling and/or protective molecule.

In Escherichia coli grown in Luria–Bertani broth, 3-mercaptopyruvate sulfurtransferase (3MST) is responsible for the bulk of endogenous H₂S generated from l-cysteine (2). Although E. coli has several known l-cysteine desulfhydrases (CDs), including O-acetylserine sulfhydrylases A and B (CysK and CysM), cystathionine β-lyases A and B (MetC and MalY), and tryptophanase (TnaA), that can, in principle, generate H₂S as a by-product of l-cysteine degradation, their contribution to H₂S production under normal growth conditions has not been established (2, 5). Because l-cysteine can be toxic to bacteria (6, 7), its intracellular level is tightly controlled. Excess l-cysteine inhibits the activity of l-serine O-acetyltransferase, a key enzyme in the l-cysteine biosynthetic pathway (8). The LysR-type transcriptional regulator, CysB, controls expression of genes involved in cysteine biosynthesis and sulfur assimilation. CysB binds the inducer, N-acetyl-l-serine (NAS), the product of a nonenzymatic rearrangement of O-acetyl-l-serine (OAS) that activates its binding to promoter DNA sequences (9). It has been shown that a high level of intracellular l-cysteine promotes the Fenton reaction (10):

$$\mathrm{L}-\mathrm{cysteine}+\mathrm{F}{\mathrm{e}}^{3+}\to \mathrm{L}-\mathrm{cystine}+\mathrm{F}{\mathrm{e}}^{2+}$$

$$\mathrm{F}{\mathrm{e}}^{2+}+{\mathrm{H}}_2{\mathrm{O}}_2\to \mathrm{F}{\mathrm{e}}^{3+}+\mathrm{\cdot OH}+\mathrm{O}{\mathrm{H}}^{-}.$$

This process is potentially toxic to the cell, because the resulting hydroxyl radicals damage nucleic acids, carbonylate proteins, and peroxidate lipids (11–13).

In our previous experiments, we showed in various bacterial species that an exogenous H₂S donor could suppress H₂O₂-mediated DNA damage (2). Here, we extend these findings to show that 3MST-mediated endogenous production of H₂S suppresses oxidative stress in E. coli by sequestering free iron required to drive the genotoxic Fenton reaction. Furthermore, we elucidate the complex interplay between 3MST and the CysB regulon that controls intracellular l-cysteine as a rate-limiting factor in H₂O₂-driven cytotoxicity.

Results

3MST-Derived H₂S Protects E. coli from H₂O₂ by Sequestering Free Iron.

To study the biochemistry of endogenous H₂S in E. coli and determine whether it is cytoprotective against ROS, we generated E. coli strains either lacking a chromosomal copy of the 3MST-encoding gene mstA (also known as sseA) or carrying it under a strong pLtetO-1 promoter (P_tet-mstA). After induced, P_tet-derived 3MST should remain at a constantly high level. We used two complementary methods to quantify the level of H₂S production by E. coli cells (Fig. 1). The first method is based on the specific reactivity of lead acetate [Pb(Ac)₂] with H₂S, resulting in a brown lead sulfide stain. The rate of staining on a Pb(Ac)₂-soaked paper strip is directly proportional to the concentration of H₂S (14). The second method uses the twister internal charge transfer (TICT)-based fluorescent probe for H₂S (15). The TICT probe is cell-permeable and allows for monitoring exogenous and endogenous H₂S in living cells. Both methods consistently show that 3MST-deficient E. coli exhibit reduced level of H₂S production, whereas P_tet-mstA cells produce much more H₂S compared with the WT (Fig. 1).

Fig. 1.

Quantitation of H₂S production by WT, 3MST-deficient (∆mstA), and 3MST-overproducing (P_tet-mstA) E. coli. (A) Representative Pb(Ac)₂-soaked paper strips show a PbS brown stain as a result of the reaction with H₂S. Strips were affixed to the inner wall of a culture tube above the level of the liquid culture of WT or mutant bacteria for 18 h. Numbers show the change in H₂S production relative to WT cells. The values are means from three independent experiments with a margin error of less than 10%. (B) Representative fluorescence images of H₂S production by live WT and mutant E. coli cells treated with the TICT-based fluorescent H₂S probe (15). (Magnification: 100×.) (C) Fluorescence intensities of WT and mutant E. coli cells in Luria–Bertani broth or Luria–Bertani broth plus 2 mM H₂O₂ treated with fluorescent H₂S probe as detected by Cytation 3 (BioTek Instruments Inc.). Values are means ± SD (n = 3). RFU, relative fluorescence unit. *P < 0.05 (Student’s t test; equal variance).

Next, we examined the sensitivity of those cells to peroxide. H₂O₂ was added to midlog phase cultures (OD₆₀₀ ∼ 0.2) at time 0, and the percentages of viable cells in the population were measured at intervals of 10, 20, and 30 min (Fig. 2A). After 20 min of treatment with 2 mM H₂O₂, the viabilities of WT and ΔmstA cells were reduced by ∼10 and 25%, respectively. P_tet-mstA cells displayed no loss of viability (Fig. 2A). Notably, the exposure of WT cells to peroxide stimulated H₂S production (Fig. 1B), indicating that cells respond to oxidative stress by stimulating the activity of 3MST.

Fig. 2.

3MST-derived H₂S protects E. coli from H₂O₂ toxicity. (A) Representative survival curves show the effect of H₂S deficiency (∆mstA) or overproduction (P_tet-mstA) on H₂O₂-mediated killing. (B) An fur mutation promotes H₂O₂ cytotoxicity in WT and ∆mstA cells but not in P_tet-mstA cells. The percentage of surviving cells was determined by counting cfu and is shown as the mean ± SD from three experiments. (C) Relative change in H₂O₂ sensitivity of WT, ∆mstA, and P_tet-mstA cells in response to Fur deficiency (∆fur). Values are means ± SD from three experiments.

H₂O₂ is only mildly genotoxic to WT K-12 E. coli, which contains little free iron (16). We, therefore, sought to promote Fenton chemistry by elevating intracellular free iron in all three strains. Ferric uptake regulator (Fur) is the master transcriptional regulator of iron uptake and homeostasis in E. coli (17, 18). For example, Fur represses a small RNA RyhB, which negatively regulates a number of iron-containing proteins in E. coli (19). Fur deletion results in a constitutive import of iron (20, 21) and hypersensitivity to oxidative DNA damage (22). Accordingly, inactivation of fur, with or without ryhB, resulted in a 40-fold increase in cell death from H₂O₂ (Fig. 2 B and C). The survivability of Δfur, or Δfur ΔryhB, cells deficient in H₂S production (ΔmstA) decreased much more drastically (∼360-fold). In contrast, Δfur, or Δfur ΔryhB, cells that overproduced H₂S (P_tet-mstA) displayed almost complete loss of susceptibility to H₂O₂ (Fig. 2 B and C), suggesting that H₂S counteracts the toxicity of H₂O₂ by sequestering the excess of free iron in Fur-deficient cells.

In support of this conclusion, we showed that the addition of FeCl₃ reduces the amount of H₂S in P_tet-mstA cells (Fig. 3A). We also observed a significant H₂S reduction in P_tet-mstA cells deleted of fur or sodA/sodB (Fig. 3A). The levels of free chelatable iron in Δfur and ΔsodA/sodB mutants and the triple Δfur ΔsodA ΔsodB mutant are ∼8- and 17-fold higher, respectively, compared with WT cells (21). Accordingly, we observed the largest decrease in detectable H₂S in P_tet-mstA cells in the Δfur ΔsodA ΔsodB mutant (Fig. 3A). Moreover, addition of an iron chelator, 2,2′-dipyridyl, fully restored the high level of H₂S in all P_tet-mstA strains (Fig. 3A). Because the inactivation of fur or sodA/sodB did not affect the level of mstA gene expression (Fig. S1), we conclude that the level of H₂S generated by 3MST is inversely proportional to the level of intracellular free iron. Taken together, these results argue that endogenous H₂S protects against H₂O₂-mediated toxicity by directly sequestrating Fe²⁺.

Fig. 3.

3MST-derived H₂S protects genomic DNA from the damaging Fenton reaction. (A) 3MST-derived H₂S sequesters intracellular iron. Representative Pb(Ac)₂-soaked paper strips show the decrease in the amount of H₂S generated in P_tet-mstA cells in response to deletion of fur or sodA sodB genes. Such deletions cause a drastic increase in the intracellular free iron content (21). Addition of 200 µM 2,2′-dipyridyl (dp), an iron chelator, restores H₂S to its original level in each case. The values (percentages) are means from three independent experiments with a margin error of less than 10%. (B) 3MST-derived H₂S renders cells less susceptible to DNA damage as evidenced by the higher H₂O₂ concentration necessary to induce the SOS response in P_tet-mstA cells. The SOS response was monitored by bioluminescence of the lux biosensor (pColD′::lux) in Δfur, Δfur ΔmstA, Δfur P_tet-mstA, and WT cells in the presence of different concentrations of H₂O₂. J/J_k indicates the induction factor in percentage compared with the maximal intensity of bioluminescence of samples in the presence of H₂O₂. Values are means ± SD from three experiments. (C) 3MST-derived H₂S renders cells less susceptible to H₂O₂-induced DNA breaks as detected by TUNEL. The graph shows the percentage of gated propidium iodide cells that are TUNEL-positive as detected by fluorescence intensity greater than that of untreated cells. Statistical evaluation (one-way ANOVA and Tukey’s post hoc test) was performed to evaluate differences in the cell population.

Fig. S1.

The relative expression of a chromosomal copy of mstA from its native promoter or Ptet-mstA in exponentially grown WT and mutant cells as measured by qRT-PCR. Values are means ± SD from three independent experiments.

3MST Is the Major CD That Protects Genomic DNA from Oxidative Damage.

Formation of double-strand breaks (DSBs) in DNA is the primary cause of bacterial cell death resulting from exposure to peroxide (23). These DSBs are the result of the toxic effects of the hydroxyl radical generated by the Fenton reaction (24). To examine whether endogenous H₂S protects bacteria from DNA damage caused by the Fenton reaction, we first examined its effect on the global response to DNA damage (SOS). We used a pColD′::lux reporter plasmid to directly monitor SOS activation in response to DNA damage (25). Fig. 3B shows the bioluminescence induction curves as a function of H₂O₂ concentrations in Δfur, Δfur ΔmstA, and Δfur P_tet-mstA cells carrying pColD′::lux. In Δfur cells, SOS induction begins at a concentration of 5 µM H₂O₂ and reaches a maximum at 80 µM followed by the decrease of bioluminescence caused by cell death. The Δfur ΔmstA mutant exhibits a maximal SOS response at the lower concentration of H₂O₂ (40 µM). In contrast, Δfur P_tet-mstA cells reach the peak of bioluminescence intensity at a much higher H₂O₂ concentration (∼1 mM), which is similar to that of the WT (Fig. 3B). These data indicate that endogenous H₂S significantly augments cellular tolerance to the Fenton reaction.

To further assess DNA damage after H₂O₂ treatment, we used an assay in which 3′-OH DNA ends were labeled with TUNEL followed by analysis by flow cytometry (Fig. 3C and Fig. S2). The percentage of TUNEL-positive cells, after gating for propidium iodide-stained cells, was significantly higher in Δfur ΔmstA than WT or Δfur P_tet-mstA cells. However, there was no significant difference in the percentages of TUNEL-positive cells between treated WT and Δfur P_tet-mstA cells. Moreover, at the 5 mM concentration of H₂O₂, the threshold of detection for TUNEL-positive cells is minimal for WT and Δfur P_tet-mstA–treated cells. These results show directly that endogenous H₂S effectively protects chromosomal DNA from H₂O₂-induced DSBs.

Fig. S2.

H₂S protects cells from H₂O₂-induced DNA damage. TUNEL analysis of H₂O₂-induced DNA damage. A–E are histograms of three biological repeat measurements of TUNEL-positive events analyzed on a FACSCalibur for each strain. Samples were treated with 5 mM H₂O₂ for 30 min at room temperature before fixation and labeling followed by analysis. The x axis (FL1-H FITC) represents the relative FITC fluorescence with respect to the untreated population, and the y axis is the number of events after gating for the propidium iodide-stained population.

The high level of resistance to oxidative stress observed in P_tet-mstA cells may not be only caused by the efficient sequestration of free iron but also, may be because of a higher rate of l-cysteine utilization via the sequential action of aspartate aminotransferase (AspC) and 3MST. l-cysteine promotes the Fenton reaction by effectively reducing Fe³⁺ to Fe²⁺ (10). Therefore, the intensive l-cysteine degradation in P_tet-mstA cells can also contribute to the suppression of the Fenton reaction.

E. coli has five known CDs in addition to 3MST, which are capable of degrading l-cysteine to pyruvate, ammonia, and sulfide. However, a quintuple mutant of ΔtnaA ΔmetC ΔcysK ΔcysM ΔmalY retains significant CD activity, which is increased in the presence of l-cysteine (5), suggesting that the major enzyme responsible for converting l-cysteine to H₂S is 3MST. Indeed, 3MST is not only responsible for the bulk of H₂S during normal growth in rich media but also, generates more H₂S under exposure to peroxide (Fig. 1C). In contrast, TnaA, which is considered to be the predominant CD (5), contributes little to the overall level of endogenous H₂S (Fig. S3A) and does not influence bacterial susceptibility to H₂O₂, irrespective of Fur (Fig. S3B).

Fig. S3.

TnaA has no effect on (A) H₂S production or (B) H₂O₂-mediated bacterial killing. The values (percentages) are means from three independent experiments with a margin error of less than 10%. Representative survival curves show the effect of TnaA deficiency (∆tnaA) or overexpression (P_tet-tnaA) on H₂O₂-mediated killing in a ∆fur background. The percentage of surviving cells was determined by counting cfu and is shown as the mean ± SD from three experiments.

Functional Interaction Between 3MST and CysB.

CysB is a master transcriptional regulator of sulfur metabolism that senses the level of endogenous l-cysteine (8). To further evaluate the impact of 3MST on endogenous l-cysteine catabolism, we used quantitative RT-PCR (qRT-PCR) to measure the expression of the CysB-dependent genes, cysK, cysP, and tau, in ΔmstA and P_tet-mstA cells. Transcription of all three genes was mildly decreased in ΔmstA cells compared with WT cells (Fig. 4A). In P_tet-mstA cells, however, cysK, cysP, and tau were induced ∼11-, 8-, and 5-fold, respectively. The induction of these genes is strictly dependent on CysB, because cysB inactivation reduced their expression to the background level (Fig. 4A). We infer that the induction of CysB-dependent genes was caused by the induction of cysB itself (Fig. 4A), which is likely to occur because of the increased l-cysteine degradation in P_tet-mstA cells. Indeed, l-cysteine is involved in feedback inhibition of serine acetyltransferase, CysE, which generates OAS, a precursor of an autoinducer for CysB, NAS (Fig. S4) (9). Accordingly, the addition of exogenous l-cysteine to the Ptet-mstA strain reduced the expression of all CysB-regulated genes to the basal level (Fig. 4A).

Fig. 4.

Functional interaction between 3MST and CysB regulon. (A) The relative expression of CysB-regulated genes in exponentially grown ΔmstA and P_tet-mstA cells was measured by qRT-PCR. The relative expression (y axis) represents the fold change of each mRNA level compared with that of the untreated cells. Values are means ± SD from four experiments. (B) Induction of CysB-regulated gene expression by H₂O₂ (2 mM, 20 min) in exponentially grown WT and ∆mstA cells as detected by qRT-PCR. The relative expression (y axis) represents the fold change of each mRNA level after treatment of the cells with H₂O₂ compared with that of the untreated cells (dashed line). Values are means ± SD from four experiments.

Fig. S4.

Interplay between H₂S generation, serine acetyltransferase activity, and cysB gene expression. The activity of serine acetyltransferase encoded by the cysE gene is subject to feedback inhibition by l-cysteine. The product of the acetyltransferase reaction is OAS, which spontaneously isomerizes to NAS. NAS acts as an inducer of the master transcriptional regulator CysB. CysB protein binds immediately upstream of the −35 region of positively regulated promoters, where in the presence of inducer, it facilitates formation of a transcription initiation complex. CysB also autoregulates its own synthesis by binding to its own promoter as a repressor. NAS stimulates CysB protein binding to sites involved in positive regulation and inhibits binding to the negatively regulated cysB promoter (9). The right-facing arrow beneath cysB indicates the direction of transcription. P, promoter.

We next examined the effect of 3MST on the CysB regulon during oxidative stress. Treatment of WT cells with 2 mM H₂O₂ for 20 min resulted in 5-, 23-, 10-, and 14-fold inductions of cysB, cysK, cysP, and tau, respectively (Fig. 4B). In contrast, the induction of CysB-regulated genes in response to H₂O₂ was completely abolished in ΔmstA cells (Fig. 4B), showing the principle role of 3MST-derived H₂S in CysB-dependent gene regulation in response to stress. Notably, the deletion or overexpression of tnaA had no effect on transcription of CysB-regulated genes (Fig. S5).

Fig. S5.

Relative expression of CysB-regulated genes in ΔtnaA and P_tet-tnaA cells. The expression was measured by qRT-PCR in exponentially growing cells. Values are means ± SD from three independent experiments.

The reciprocal communication between CysB and 3MST is further evident from the requirement of CysB for H₂S-mediated protection against oxidative stress. Inactivation of cysB increased the sensitivity of ∆fur cells to H₂O₂, which cannot be suppressed by P_tet-mstA (Fig. S6). Moreover, inactivation of cysB almost completely abolished H₂S generation by P_tet-mstA cells (Fig. 5A). We suggest that, without CysB, the transport of l-cysteine into the cell is abrogated, hence the inability of 3MST to generate H₂S and protect against oxidative stress.

Fig. S6.

Deletion of cysB abolishes the protective effect of H₂S against H₂O₂. Overnight cultures of Δfur (squares), Δfur ΔmstA (triangles), and Δfur P_tet-mstA (circles) in (A) cysB+ and (B) ΔcysB backgrounds were inoculated in Luria–Bertani broth liquid medium and grown to OD ∼ 0.2 followed by addition of 2 mM H₂O₂ (black) or water (white). Cells were grown in triplicate at 37 °C with aeration using a Bioscreen C automated growth analysis system. The curves represent the averaged values from three parallel experiments with a margin of error of less than 5%.

Fig. 5.

Interdependence between 3MST activity and l-cysteine/cystine import. Constitutive expression of the TcyP transporter suppresses the negative effect of cysB deletion on H₂S production in (A) P_tet-mstA cells or (B) ∆mstA and WT cells as detected by the Pb(Ac)₂ assay. Representative panels show mean values (percentages) from three independent experiments with a margin error of less than 10%. (C) A model of H₂S-mediated defense against oxidative stress in E. coli. A fraction of exogenous H₂O₂ reacts with l-cysteine in the periplasm to form l-cystine and H₂O. This reaction leads to a decrease in the intracellular content of l-cysteine with a subsequent relief of autoregulation of cysB and activation of CysB-dependent genes, including tcyP, which is responsible for transport of l-cystine into the cell. Overflow of cystine/cysteine flux results in increased mstA-dependent generation of H₂S, which sequesters free iron to prevent the Fenton reaction and formation of damaging hydroxyl radicals.

To test this hypothesis, we placed the chromosomal copy of the major l-cystine importer, tcyP (26, 27), under the strong Ptet promoter. TcyP is normally under the positive control of CysB (28). Ptet-tcyP fully restored 3MST-dependent H₂S production in Ptet-mstA cells (Fig. 5A). Moreover, Ptet-tcyP increased H₂S production in cysB(−) or (+) cells carrying mstA under its native promoter (Fig. 5B). Because the deletion of mstA in Ptet-tcyP cells abolishes H₂S production, we conclude that 3MST is the sole source of H₂S in E. coli grown in Luria–Bertani broth. These results argue that, under conditions of cystine overflow, the AspC-3MST system generates a sufficient amount of H₂S to render cells resistant to oxidative stress. To maintain such a protective level of H₂S under oxidative stress, the enhanced influx of l-cysteine must occur. Accordingly, the expression of tcyP is strongly induced in response to H₂O₂ treatment (Fig. S7). Moreover, this induction is strictly dependent of 3MST activity: deletion of mstA abolishes tcyP induction, whereas Ptet-mstA increases it (Fig. S7).

Fig. S7.

Induction of tcyP gene expression by H₂O₂ in exponentially grown WT, ∆mstA, and P_tet-mstA cells as measured by qRT-PCR. The relative expression (y axis) represents the fold change of each mRNA level after treatment of the cells with 2 mM H₂O₂ for 20 min compared with that of the untreated cells (dotted line). Values are means ± SD from three independent experiments.

Discussion

The purpose of this work is to explain the mechanism of H₂S-mediated protection against oxidative stress and establish the biochemical pathway of H₂S production in response to stress in E. coli. The results determine that the AspC-3MST pathway is the principle source of H₂S in E. coli grown in rich medium containing cysteine (Fig. S4). It has been assumed that TnaA could be the major CD and potential generator of H₂S in E. coli (5). However, our previous work showed that inactivation of TnaA (ΔtnaA) or other known desulfhydrases (ΔmetC, ΔcysK, ΔcysM, and ΔmalY) does not significantly alter the level of endogenous H₂S (2). Here, we provide independent support for this conclusion and show that the inactivation or overexpression of TnaA does not function at all in H₂S-mediated protection against oxidative stress (Fig. S3). Rather, 3MST is central.

The protective function of 3MST becomes most apparent in Fur-deficient cells, in which the level of intracellular iron (Fe²⁺) substantially increased (22). The ∆mstA ∆fur double mutant exhibited a 360-fold increase in sensitivity to H₂O₂ compared with its ∆mstA fur⁺ counterpart (Fig. 2), which showed an ∼10-fold increase in peroxide sensitivity compared with the ∆fur mutant. This sensitivity correlates well with the dramatic increase in genomic DNA DSBs (Fig. 3C and Fig. S2). Remarkably, endogenous overproduction of H₂S from the chromosomal Ptet-mstA completely protects Fur-deficient cells from H₂O₂ toxicity and DNA damage. Furthermore, we found that the level of H₂S in P_tet-mstA cells is reduced in Δfur or ΔsodA ΔsodB cells but can be restored after addition of the FE²⁺ chelator, 2,2′-dipyridyl (Fig. 3A). These data imply that 3MST renders E. coli resistant to oxidative stress via H₂S-mediated sequestration of Fe²⁺, thereby diminishing the genotoxic Fenton reaction (Fig. 5C).

Because the amino acids in Luria–Bertani broth are the main carbon source (29), we postulate that Luria–Bertani broth-derived l-cystine/cysteine is the principle substrate for H₂S production by AspC-3MST. Indeed, the deletion of cysB abolishes the generation of H₂S in P_tet-mstA cells. CysB positively regulates not only the genes responsible for l-cysteine biosynthesis but also, tcyP and tcyJ, which encode the two l-cystine transporters, the symporter TcyP and the ATP binding cassette importer TcyJ, respectively (27). Therefore, the inability of the Ptet-mstA ∆cysB mutant to generate H₂S can be caused by reduced production of endogenous l-cysteine, disruption of l-cystine import from the Luria–Bertani broth medium, or both. We found that the introduction of the constitutively active form of tcyP (Ptet-tcyP) (Fig. 5A), but not tcyJ (Ptet-tcyJ) (Fig. S8), fully restores the generation of H₂S in CysB-deficient Ptet-mstA cells. Remarkably, we found that the constitutive expression of tcyP also leads to overproduction of H₂S in cells with native expression of mstA (Fig. 5B). Thus, the main source of H₂S generated by 3MST is l-cystine/cysteine imported from the Luria–Bertani broth medium by the TcyP transporter (Fig. 5C). This conclusion is consistent with the observation that, unlike TcyJ, TcyP functions predominantly as a nutrient importer under normal growth conditions (26).

Fig. S8.

Constitutive expression of the TcyJ transporter does not suppress the negative effect of cysB deletion on H₂S generation in (A) P_tet-mstA or (B) WT, ΔmstA, or ΔcysB cells. Pb(Ac)₂-soaked paper strips show a PbS brown stain as a result of reaction with H₂S. Strips were affixed to the inner wall of a culture tube above the level of the liquid culture of bacteria for 18 h. The values (percentages) are means from three experiments with a margin error of less than 10%.

Our results also reveal the reciprocal interaction between 3MST and the CysB regulon under normal growth conditions and during oxidative stress. The high level of 3MST expression in Ptet-mstA cells resulted in cysB induction and its target genes (cysK, cysP, and tau), whereas in the absence of 3MST, the expression of all CysB-regulated genes was diminished (Fig. 4A). Remarkably, 3MST deficiency also abolished H₂O₂-mediated induction of CysB-dependent genes (Fig. 4B). It has been reported that at least three such genes (cysK, cysP, and tcyJ) are highly up-regulated in response to H₂O₂ in an OxyR-independent manner (26, 30). The mechanism of such an induction remains unknown. Our results suggest the following model, which explains the interplay between oxidative stress, activation of the CysB regulon, and 3MST-dependent generation of H₂S (Fig. 5C). The sulfhydryl group of l-cysteine reacts with H₂O₂ in the periplasm to yield l-cystine (26). This reaction lowers the intracellular level of l-cysteine leading to the induction of the CysB regulon, including the TcyP transporter, thereby boosting the l-cystine/cysteine influx into the cytoplasm. The increased flow of l-cysteine stimulates H₂S production by the AspC-3MST pathway, leading to sequestration of Fe²⁺ and suppression of the Fenton reaction (Fig. 5C). Inactivation of 3MST halts the conversion of l-cysteine to H₂S, leading to accumulation of intracellular l-cysteine, thereby preventing H₂O₂-dependent induction of CysB-regulated genes and fueling the genotoxic Fenton reaction.

Understanding the mechanism of H₂S-mediated protection against ROS has important implications for bacterial resistance to antibiotics (31, 32). Pharmacological inhibition of bacterial H₂S production may facilitate rapid bacterial killing, which would not only widen the therapeutic window for many classes of bactericidal antibiotics but also, diminish the rate at which bacteria acquire resistance to such antibiotics (33).

Materials and Methods

Strains and Growth Conditions.

All E. coli strains used in this work are listed in Table S1. BW25113 and its derivatives (single-gene deletion mutants) were obtained from the E. coli Keio Knockout Collection (Thermo Scientific) (34). Details of strain constructions are described in SI Materials and Methods. P1 transduction was used to introduce mutations into new strains (35). When necessary, Cam or Kan drug resistance markers were excised from strains using the flippase activity of pCP20 followed by loss of the plasmid at nonpermissive temperature (36). All mutations were verified by PCR and gel analysis. DNA manipulations and the transformation of E. coli strains were performed according to standard methods (37). Luria–Bertani broth complete medium was used for the general cultivation of E. coli. When appropriate, antibiotics were added at 40 µg/mL (for kanamycin), 30 µg/mL (for chloramphenicol), and 100 µg/mL (for ampicillin). For solid medium, 1.5% agar was added.

Table S1.

Strains used in this study

Strain name	Genotype	Source
MG1655	F− WT Escherichia coli	Ref. 2
AM3007	As MG1655 plus ∆mstA	Ref. 2
AM3009	As MG1655 plus Ptet-mstA	Ref. 2
BW25113	F− Δ(araD-araB)567, lacZ4787(::rrnB-3), λ−, rph-1, Δ(rhaD-rhaB)568, hsdR514 with pKD46-(Ts)	Ref. 36
JW0669	As BW25113 plus ∆fur::kan	Keio collection (34)
JW3914	As BW25113 plus ∆katG::kan	Keio collection (34)
JW3879	As BW25113 plus ∆sodA::kan	Keio collection (34)
JW1684	As BW25113 plus ∆sodB::kan	Keio collection (34)
JW1267	As BW25113 plus ∆cysB::kan	Keio collection (34)
JW3686	As BW25113 plus ∆tnaA::kan	Keio collection (34)
AM3011	As MG1655 plus ∆fur::kan	P1(JW0669) × MG1655
AM3012	As AM3011 ∆fur	This work
AM3015	As AM3007∆mstA plus ∆fur::kan	P1(JW0669) × AM3007
AM3017	As AM3007∆mstA plus ∆fur	This work
AM3018	As AM3009 Ptet-mstA plus ∆fur::kan	P1(JW0669) × AM3009
AM3019	As AM3009 Ptet-mstA plus ∆fur	This work
AM3022	As AM3012∆fur plus ∆katG::kan	P1(JW3914) × AM3012
AM3025	As AM3022∆fur plus ∆katG	This work
AM3028	As AM3017∆mstA ∆fur plus ∆katG::kan	P1(JW3914) × AM3017
AM3030	As AM3028∆mstA ∆fur plus ∆katG	This work
AM3034	As AM3019 Ptet-mstA∆fur plus ∆katG::kan	P1(JW3914) × AM3019
AM3035	As AM3019 Ptet-mstA∆fur plus ∆katG	This work
AM3038	As AM3009 Ptet-mstA plus ∆sodA::kan	P1(JW3879) × AM3009
AM3039	As AM3038 Ptet-mstA plus ∆sodA	This work
AM3042	As AM3039 Ptet-mstA ∆sodA plus ∆sodB::kan	P1(JW1684) × AM3039
AM3043	As AM3039 Ptet-mstA ∆sodA plus ∆sodA	This work
AM3047	As AM3043 Ptet-mstA ∆sodA ∆sodA plus ∆fur::kan	P1(JW0669) × AM3043
AM3048	As AM3043 Ptet-mstA ∆sodA ∆sodA plus ∆fur	This work
AM3050	As MG1655 plus ∆tnaA::kan	P1(JW3686) × MG1655
AM3051	As MG1655 plus Ptet-tnaA	This work
AM3052	As MG1655 plus ∆cysB::kan	P1(JW1267) × MG1655
AM3057	As AM3007 ∆mstA plus ∆cysB::kan	P1(JW1267) × AM3007
AM3059	As AM3009 Ptet-mstA plus ∆cysB::kan	P1(JW1267) × AM3009
AM3062	As MG1655 plus Ptet-tcyJ	This work
AM3064	As MG1655 plus Ptet-tcyP	This work
AM3065	As AM3007 ∆mstA plus Ptet-tcyJ	This work
AM3067	As AM3007 ∆mstA plus Ptet-tcyP	This work
AM3069	As AM3009 Ptet-mstA plus Ptet-tcyJ	This work
AM3070	As AM3009 Ptet-mstA plus Ptet-tcyP	This work

Generation of Growth Curves.

Growth curves were obtained on a Bioscreen C automated growth analysis system. Subcultures of specified strains were grown overnight at 37 °C, diluted in fresh medium at 1:100, inoculated into honeycomb wells in triplicate, and grown at 37 °C with maximum shaking on the platform of the Bioscreen C instrument. When the cultures reached an OD₆₀₀ of 0.2, cells were treated with H₂O₂ (2 mM) and incubated at 37 °C for 10 h. OD₆₀₀ values were recorded automatically at specified times, and the mean value of the triplicate cultures was plotted.

Generation of Survival Curves.

Overnight cultures were inoculated into Luria–Bertani broth and grown at 37 °C to ∼2 × 10⁷ cells per 1 mL. Cells were then treated with H₂O₂ (2 mM) and after 10 or 20 min of incubation, samples were diluted, plated on Luria–Bertani broth agar, and incubated at 37 °C for 16–18 h. Cell survival was determined by counting cfu and is shown as the mean value ± SD from three independent experiments.

H₂S Detection.

To monitor H₂S production, we used a Pb(Ac)₂ detection method (14) and the TICT-based fluorescent H₂S probe (BH-HS) (15). Overnight cultures were diluted 1:500 in Luria–Bertani broth and incubated at 37 °C with aeration (250 rpm) for 18–20 or 3–4 h for Pb(Ac)₂ or BH-HS, respectively. Before incubation, the paper strips saturated with 2% Pb(Ac)₂ were affixed to the inner wall of a cultural tube above the level of the liquid culture of WT or mutant bacteria. Stained paper strips were scanned and quantified with an Alpha Imager (Imgen Technologies). BH-HS (5 μM) was added to liquid bacterial culture, and after 40 min, the aliquots were taken for fluorescent microscopy (API DeltaVision PersonalDV system with Olympus IX-71 inverted microscope base). Images were taken with an Olympus PlanApo N 60×/1.42 oil lens. A Cytation 3 (BioTek Instruments Inc.) was used to quantitate fluorescence. The results were normalized according to the ODs.

Measurement of Luminescent Reaction of Lux Biosensors.

The SOS response was examined using a pColD′::lux hybrid plasmid (38), a derivative of the pDEW201 vector containing luxCDABE from Photorhabdus luminescens under the control of the LexA-regulated Pcda promoter (25). Overnight cultures of strains containing the pColD′::lux plasmid were diluted to a concentration of 10⁷ cells per 1 mL in fresh Luria–Bertani broth medium and grown under aeration at 30 °C until the early exponential growth phase; 200-μL aliquots were transferred into special cuvettes, one of which served as a control (4 mL distilled water was added to the control cuvette), and 4 mL peroxide was introduced at various concentrations into the other cuvettes. Samples of lux biosensors thus prepared were placed in front of a photomultiplier in an LMAO1 luminometer (Beckman), and the intensity of bioluminescence of the cell suspension was measured at certain times. The samples were incubated at room temperature. The bioluminescence intensity was determined according to ref. 39.

RNA Extraction and qRT-PCR.

E. coli K-12 MG1655 cells were grown until OD₆₀₀ of 0.6, and total RNA was extracted using the RNeasy Mini Kit (QIAGEN) according to the manufacturer’s protocol. All RNA samples were treated with DNaseI (Fermentas); 500 ng total RNA was reverse-transcribed with 100 U SuperScript III enzyme from the First-Strand Synthesis Kit for RT-PCR (Invitrogen) according to the manufacturer’s protocol in the presence of appropriate gene-specific primers (Table S2). One microliter reverse transcription reaction was used as the template for real-time PCR. The gene def encoding peptide deformylase was used for normalization. Each real-time PCR mixture (25 μL) contained 10 μL SYBR Green I PCR Master Mix (Syntol), 12 μL nuclease-free H₂O, 1 μL 10 μM forward primer, 1 μL 10 μM reverse primer, and 1 μL cDNA template. Amplifications were carried out using the DTlite S1 CyclerSystem (DNA Technology). Reaction products were analyzed using 2% agarose electrophoresis to confirm that the detected signals originated from products of expected lengths. Each qRT-PCR was performed at least in triplicate, and average data are reported. Error bars correspond to the SD.

Table S2.

Oligonucleotides used in this study

Name	Sequence
cysB-f	5′ CAACAACTTCGCTATATTGTTGAG 3′
cysB-r	5′ CATCGACTTTCGACAGGACTTCG 3′
cysP-f	5′ GTTAACTTACTGAAAAAGAACTCAC 3′
cysP-r	5′ CACCTGGTTATAAGTGACAACGTCG 3′
cysK-f	5′ GATAACTCGCTGACTATCGGTCACAC 3′
cysK-r	5′ CCAGTGCAATCCCGGTATTACCGCTG 3′
tauA-f	5′ GCAATTTCATCGCGTAACACACTTC 3′
tauA-r	5′ CGAGGTTGCCGATTTGCACGTCGC 3′
cysE-f	5′ CGTGTGAAGAACTGGAAATTGTC 3′
cysE-r	5′ GTACGCACCGCCTGAATATCAC 3′
mstA-RTf	5′ CTCAGACAGCCAAATCAGCCTC 3′
mstA-RTr	5′ GTCATCAATATGTTCGGCGAGC 3′
Def-4	5′ ACTTCTTCTACCGGTTTAGCA 3′
Def-11	5′ AGATTTATGTCAGTTTTGCAAGT 3′

TUNEL Assay.

Cells were grown until OD₆₀₀ of 0.4, and 1-mL aliquots were treated with 5 mM H₂O₂ for 30 min. Cells were fixed and labeled using a slightly modified protocol for the Apo-Direct TUNEL assay kit (EMD Millipore). Briefly, treated cells were harvested, washed, and resuspended in 1 mL cold 4% paraformaldehyde, and then, they were incubated on ice. After 1 h, cells were centrifuged, washed, and resuspended in 70% ethanol overnight at −20 °C. The next day, cells were centrifuged, washed, and resuspended in 50 μL TUNEL reaction mix for 2 h at 37 °C. After the labeling reaction was stopped, the cells were counterstained with propidium iodide/RNase A and analyzed by flow cytometry on the FACSCalibur.

SI Materials and Methods

To construct the tnaA, tcyJ, and tcyP overexpression strains, we substituted the native promoter of these genes with P_Ltet-O1 (40). Briefly, the P_Ltet-O1-attL-Cm^R-attR cassette integrated in the AM3009 strain into the mstA gene (2) was amplified with primers 5′-tttgcccttctgtagccatcaccagagccaaaccgcgctcaagttagtataaaaaagct-3′ and 5′-gagatgtttaaagttttccattacataatccttcatggtacctttctcctctttaatga-3′ (for tnaA gene), 5′-cctccagcctgccttcttctgatatatattaaataacgctcaagttagtataaaaaagct-3′ and 5′-catcaatgcctgacgtcccagatgtgctaatttcatggtacctttctcctctttaatga-3′ (for tcyJ gene), and 5′-tgttaatcttgcgccaacactatgactgctacgcagcgctcaagttagtataaaaaagct-3′ and 5′-gaacaccacgatgttcgcaattaatggaaagttcatggtacctttctcctctttaatga-3′ (for tcyP gene). The first primers contain the upstream region of the tnaA, tcyJ, and tcyP genes and the sequence of attR, whereas the second primers contain the coding region of the corresponding genes and sequence of P_Ltet-O1. The PCR fragments were transformed into MG1655 containing pKD46 (35). Cm^R clones were tested for the presence of the P_Ltet-_O1-attL-Cm^R-attR cassette by PCR with primers 5′-cccgaacgattgtgattcgatt-3′ and 5′-ctcttcacgataagcgcgag-3′ (for tnaA gene), 5′-ccaccaccagcaccaacaa-3′ and 5′-atgttttgccagctgttggg-3′ (for tcyJ gene), and 5′-ctggaataagcaattccatttg-3′and 5′-gttaccaacgatgttaaaccac-3′ (for tcyP gene).